SmartMatch requires that photos have ‘texture’ (i.e., contrasting image features) so that minute features in each photo can be matched between photos. The feature detection algorithm detects features that it thinks are significant, based on the distinct feature’s position, scale, shape etc. in the image. You may notice that SmartPoints don’t get marked in necessarily predictable locations. The algorithm’s sophisticated search function will detect features based on a series of checks to ensure the highest reliability of a feature match between photos.
Good texture and patterns are brick/stone buildings, rock walls, stockpiles, quarries, archaeological digs, statues, skin, trees, wood grain, etc.. Reflective surfaces can be confusing because the glare or reflections are not stationary and move relative to the camera. If a texture is not constant between photos, the matching can become unreliable. If the texture is too uniform (e.g. a smooth painted surface over a large portion of the image), distinct features cannot be adequately detected and will cause problems for SmartMatch. Transparent materials will also cause problems due to glare, reflectivity, and distortion of features.
Object/scene lighting can have an effect on texture appearance. Avoid using lighting in the scene that causes glare, reflections or large deep shadows on the surface. This is especially true of shiny surfaces. Matte surfaces that do not show highlights from directed light sources are less prone to this problem and some side lighting and shadows are acceptable. Usually it is not desirable to use a flash mounted on the camera because the shadows would move on the object between photos. Other similar light sources (including a bright sun) should not move between photographs. In the case of sunshine and outdoor scenes do not take photos hours or days apart, so that these types of lighting issues are minimized.
Video frame extraction is also available for SmartMatch projects. When selecting images for your project, select a supported video file; you will then be prompted to automatically or manually extract frames from the video. See Video Frame Extraction.
Note: SmartMatch projects will not work with a stationary camera and the object on a turntable, because the background scene will change relative to the foreground object, and this will confuse the matching algorithms.